The future is messy

The Dutch elections are not unusual in creating a fragmented parliament

Most people who followed yesterday’s Dutch elections did so because of a one-man party which won 13% of the vote.

It clearly matters that Wilders’ PVV failed to improve on its best performance so far, and that it did not finish as the largest party in the Tweede Kamer, as polls earlier in the year had suggested it might.

Yet what probably matters more is the fragmentation of the Dutch party system.

Rather than top-three (or top-two, or top-four) share, political scientists use a measure of fragmentation called the effective number of parties.

To calculate the effective number of parties, you take each parties’ share of the vote, multiply it by itself, add up the squares, and divide one by the result.

If you have two parties with 0.5 (50%) of the vote each, that’s 1/(0.5² + 0.5²) = 2.

If you have four parties with 0.25 (25%) of the vote each, that’s 1/(0.25² + 0.2² +0.25² +0.25²) = 4, and so on.

The value for yesterday’s Dutch election is 8.6, which puts the Dutch election in the top 5% of parliamentary elections, according to stats from ParlGov.

Although the Dutch are extreme, other countries are becoming more like the Netherlands. I’ve plotted below the effective number of (electoral) parties for parliamentary democracies, according to whether or not they were:

  • democracies before 1960(Denmark, Australia, New Zealand, Belgium, Canada, Sweden, Switzerland, Norway, Iceland, Ireland, Finland, France, Netherlands, Japan, United Kingdom, Luxembourg, Austria, Israel, Italy, Germany, Malta)
  • democracies after 1960 but before 1989(Greece, Portugal, Spain, Cyprus)
  • democracies after 1989 (Croatia, Estonia, Hungary, Romania, Bulgaria, Czech Republic, Lithuania, Latvia, Slovenia, Slovakia, Poland)

The trend lines are local regression lines. Countries with an effective number of parties greater than eight are labelled.

Because the Netherlands has a hyperproportional electoral system, the effective number of electoral parties — calculated on the basis of parties’ vote shares — is broadly similar to the effective number of legislative parties, calculated on the basis of parties’ seat shares. The Greek elections of May 2012 were very fragmented at the electoral level, but the electoral system produced a less fragmented parliament.

If we’re prepared to concentrate on the effective number of electoral parties, we’ll see how the trend line for first-wave democracies (in blue, calculated excluding the Dutch result) has already caught up with the trend for the third-wave democracies (in green). Second-wave democracies (a rather amorphous category) have also seen rapid increases in fragmentation.

I don’t wish to suggest that, in the future, every parliament will have space for fifteen parties. But so far there is a trend towards increased fragmentation (which must necessarily damage existing “established” party families like the family of social democrats). Strategies for managing that fragmentation — without furthering the impression that governmental politics is performed by a small set of cartelised parties — are going to become much more important.

Note: Replication code is on GitHub.